Mapping Brooklyn Borough to visualize the areas of interest to find out the food desert

As per the USDA, an area can be classified as food desert if it meets the following two conditions:

  1. Poverty Rate is greater than or equal to 20% or a household with income less than the federal poverty level of USD 17,050 for a family of four in 2000

  2. At least 500 people or 33% of the population is located more than 1 mile (Urban) and 10 miles (Rural) from the nearest supermarket or grocery store

Insights found during Visualization

After visualizing the Food Access Research Atlas 2019 using GIS tool of Geopandas, it was found that in Brooklyn, there are multiple census tracts having greater than or equal to 20% poverty rate which meets condition one but not a single tract meets condition 2 which technically suggests that Brooklyn do not have any Food Desert as per the official definition given by US Department of Agriculture (USDA).

However, Based on multiple articles published by community bloggers, it is found that Central Brooklyn is one of two worst food desert with neighborhoods such as Ocean Hill, Brownsville, and Bedford-Stuyvesant being the lowest income members.

Few such sources are:

  1. https://neighborstogether.org/our-community/
  2. https://www.thecity.nyc/2022/07/14/eric-adams-food-apartheid-bed-stuy-residents-affordable-grocery/
  3. https://www.cbsnews.com/newyork/news/food-insecurity-remains-big-problem-in-more-than-2-dozen-neighborhoods-in-new-york-city/

The following codes will take a look into variables from different sources and map them over the borough of Brooklyn

Sources used are:

  1. New York State Map : Uploaded on Github under Data Collection folder as cb_2019_36_tract_500k.shx.
  2. Food Access Research Atlas 2019 : Uploaded on Github under Data Collection folder as Kings_data.csv

  3. Grocery Store Database 2017: https://nanda.isr.umich.edu/project/grocery-stores/

  4. Socio-Economic Database 2016 https://nanda.isr.umich.edu/project/socioeconomic-status-and-demographic-characteristics/
  5. Eating & Drinking Database 2017:https://nanda.isr.umich.edu/project/eating-and-drinking-places/
  6. Convenience Stores 2017:https://nanda.isr.umich.edu/project/liquor-tobacco-and-convenience-stores/

Database 1: New York State Map

Database 2: Food Access Research Atlas 2019 - Kings County

Visualizing Low Income and Low Access Census Tracts which are 1/2 mile from nearest supermarket

Database 3: Grocery Store 2003 - 2017

This database contains the following

  1. 445110: supermarkets and grocery stores, excluding convenience stores.
  2. 4452: all specialty food stores. These include meat, fish, and seafood markets, produce markets, baked goods stores, and spice stores.
  3. 452311: warehouse clubs and supercenters, which sell fresh food alongside canned and packaged foods and other types of merchandise (such as clothing and furniture).

2017 Grocery data was used as a proxy to 2019 Food Access Research Atlas data

Database 4: Socio-Economic Data

Database 5: Eating & Drinking Data

This database contains the following

  1. 7225: all restaurants and eating places.
  2. 722511: full-service restaurants where patrons order and are served while seated.
  3. 722513: limited service restaurants, e.g. fast-food restaurants.
  4. 722515: snack and nonalcoholic beverage bars, such as coffee shops, doughnut or bagel shops, and ice cream parlors.
  5. 722410: drinking places (alcoholic beverages), such as bars, taverns, and cocktail lounges.

2017 Eating & Drinking Data is used as a proxy to 2019 Food Access Research Atlas data

Database 6: Convenience Stores. Liqour and Tobacco stores

This database contains the following

  1. 4453: beer, wine, and liquor stores (also known as package stores).
  2. 453991: cigar, cigarette, and tobacco stores, excluding stores that sell electronic cigarettes.
  3. 445120: convenience stores without gas stations. These are defined by NAICS as “establishments…primarily engaged in retailing a limited line of goods that generally includes milk, bread, soda, and snacks.”
  4. 447110: gas stations with convenience stores

2017 Convenience stores, Liqour and Tobacco database is used as a proxy to 2019 Food Access Research Atlas data

Database 7: Dollar Stores

This database contains the following

  1. 452319: Dollar Stores

Final merged dataset

Transforming few of the variables from string to float values

Visualizing how population having Income less than 40K is distributed across Brooklyn

Visualizing how population having Income between 40K and 75K is distributed across Brooklyn

Visualizing how population accessing SNAP Benefits is distributed across Brooklyn

Visualizing how the population having Education below high school/diploma is distributed across Brooklyn

Plotting SNAP Benefits vs. Income less than 40K

Experiment 1: Deriving Food Index for each neighborhood

Calculating Food Index based on Poverty, # of Supermarkets, # of Fast Food Restaurants and # of Coffee Shops

a) Normalizing Poverty Rate, # of Supermarkets, # of Fast Food Restaurants, # of Coffee Shops

b) Applying weights to Food Outlets

c) Calculating Food Index for each Neighborhood

Creating subset of final merged dataset

K-Means Clustering

Inferences:

  1. The clusters seems to show spatial autocorrelation but with poor Silhouette Scores
  2. It is somewhat capturing the extreme food desert areas but not entirely

DB Scan

Insights derived from DB Scan Clusters

  1. The Food Index corresponds with SNAP Benefits, Poverty Rates, Median Income, Education and People earning less than 40 K a year

Next Steps

1. Identifying the policies implemented since 2009 and using sliders to see if the Food Index changes over time

2. We need to have Poverty Rates at Census Tract Levels from 2009 to 2024

3. Aid in recommending new policies based on changing/updating the variables